\begin{table}[htbp]\centering
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption{US (HRS): Time trend and unemployment rate regression results}
\begin{tabular*}{\hsize}{@{\hskip\tabcolsep\extracolsep\fill}l*{3}{c}}
\toprule
            &\multicolumn{1}{c}{(1)}         &\multicolumn{1}{c}{(2)}         &\multicolumn{1}{c}{(3)}         \\
            &         AME         &         AME         &         AME         \\
            &                     &                     &                     \\
\midrule
Unemployment rate*100&        .898\sym{***}&        .837\sym{***}&        .943\sym{***}\\
            &     (.0539)         &     (.0705)         &      (.041)         \\
t           &        .964\sym{*}  &        .992\sym{**} &                     \\
            &      (.316)         &      (.273)         &                     \\
t^{2}       &                     &       -.647         &                     \\
            &                     &       (.47)         &                     \\
Linear spline (pre-2008)&                     &                     &        1.35\sym{***}\\
            &                     &                     &      (.232)         \\
Linear spline (post-2008)&                     &                     &       .0655\sym{*}  \\
            &                     &                     &     (.0203)         \\
\midrule
\(N\)       &          12         &          12         &          12         \\
\(R^{2}\)   &       0.883         &       0.911         &       0.940         \\
\tabnotes{4}{This table displays the results of regressions of the average marginal effects of the year dummies (with controls) on the unemployment rate and various functional forms of time using the HRS data. In the quadratic functional form, \textit{t} is re-parameterized to be equal to \textit{year} - 2016. Standard errors in parentheses. *p < 0.05, **p < 0.01 ***p < 0.001}
